J Manipulative Physiol Ther. 2004 (Jan); 27 (1): 1–15 ~ FULL TEXT
Christopher J Colloca, DC, Tony S Keller, PhD, Robert Gunzburg, MD, PhD
State of the Art Chiropractic Center,
Phoenix, AZ 85044, USA.
OBJECTIVE: The purpose of this study was to quantify in vivo vertebral motions and neurophysiological responses during spinal manipulation.
METHODS: Nine patients undergoing lumbar decompression surgery participated in this study. Spinal manipulative thrusts (SMTs) ( approximately 5 ms; 30 N [Sham], 88 N, 117 N, and 150 N [max]) were administered to lumbar spine facet joints (FJs) and spinous processes (SPs) adjacent to an intraosseous pin with an attached triaxial accelerometer and bipolar electrodes cradled around the S1 spinal nerve roots. Peak baseline amplitude compound action potential (CAP) response and peak-peak amplitude axial (AX), posterior-anterior (PA), and medial-lateral (ML) acceleration time and displacement time responses were computed for each SMT. Within-subject statistical analyses of the effects of contact point and force magnitude on vertebral displacements and CAP responses were performed.
RESULTS: SMTs (>/= 88 N) resulted in significantly greater peak-to-peak ML, PA, and AX vertebral displacements compared with sham thrusts (P <.002). SMTs delivered to the FJs resulted in approximately 3–fold greater ML motions compared with SPs (P <.001). SMTs over the SPs resulted in significantly greater AX displacements compared with SMTs applied to the FJs (P <.05). Seventy-five percent of SMTs resulted in positive CAP responses with a mean latency of 12.0 ms. Collectively, the magnitude of the CAP responses was significantly greater for max setting SMTs compared with sham (P <.01).
CONCLUSIONS: Impulsive SMTs in human subjects were found to stimulate spinal nerve root responses that were temporally related to the onset of vertebral motion. Further work, including examination of the frequency and force duration dependency of SMT, is necessary to elucidate the clinical relevance of enhanced or absent CAP responses in patients.
From the Full-Text Article:
Because spinal manipulation (SM) is a mechanical intervention, it is inherently logical to assume that its mechanisms of therapeutic benefit may lie in the mechanical properties of the applied force (mechanical mechanisms), the body's response to such force (mechanical or physiologic mechanisms), or a combination of these and other factors. Basic science research, including biomechanical and neurophysiological investigations of the body's response to SM, therefore, should assist researchers, educators, and clinicians to understand the mechanisms of SM, to more fully develop SM techniques, to better train clinicians, and ultimately attempt to minimize risks while achieving better results with patients.
From a biomechanical perspective, human cadaver and in vivo studies have characterized the forces and force-time histories associated with various spinal manipulation techniques. [1–9] These studies provide important information concerning the forces and loading history transmitted to patients. The posterior-anterior (PA) stiffness or PA load-displacement response of the prone-lying subject during SM has also been investigated using static or low-frequency indentation types of techniques, including mobilization and other physiotherapy simulation devices. [10–15] These studies indicate that the thoracolumbar spine has a quasi-static PA structural stiffness of approximately 15 to 30 N/mm at loads up to about 100 N. While stiffness measurements quantify the force-displacement response of the area under test (vertebrae, disks, and associated soft tissues), such measurements cannot easily distinguish the contribution and/or displacement of individual vertebral components.  To precisely quantify relative and absolute movements of individual vertebrae or motion segments in response to dynamic forces, it is necessary to measure displacements, velocities, or accelerations using transducers fixed to intraosseous pins rigidly attached to the spine. Due to the invasiveness of such procedures, however, these techniques are generally limited to studies of human cadavers [17, 18] or animals. [19, 20] Indeed, research of this nature in living humans is very rare. 
In 1994, Nathan and Keller  quantified the sagittal plane, intersegmental motion response and stiffness of the thoracolumbar spine of human subjects during mechanical-force, manually-assisted (MFMA) short lever spinal manipulative thrusts (SMTs). In their study, forces were delivered to the spinous processes of the thoracolumbar spine using a chiropractic adjusting instrument equipped with a load cell and accelerometer. The motion response of adjacent lumbar vertebrae was quantified using an intervertebral motion device (IMD)  attached directly to intraosseous pins fixed to lumbar spinous processes. They found that the peak-to-peak amplitude of intervertebral or intersegmental motions were up to 6–fold greater when the short duration (< 5 milliseconds) SMTs were delivered closer to the IMD measurement site. In response to the same force amplitude, differences in intervertebral acceleration time and displacement time histories were also noted among the 3 subjects examined in this study (1 normal subject and 2 subjects consulting for surgery). The study by Nathan and Keller  was limited to a single force amplitude PA thrust applied over the spinous processes, and only the relative movements of 2 adjacent vertebrae (intersegmental motion) could be determined. To our knowledge, there are no data in the literature that characterize the segmental and intersegmental motion responses of the spine to varying force amplitudes and contact points in living subjects.
From a neurophysiological perspective, the presence of mechanosensitive and nociceptive afferent fibers in spinal tissues (disk, facet, ligaments, and muscles) [24–28] and the subsequent neurophysiological research demonstrating the role of such afferent stimulation in pain production [29–31] and coordinated neuromuscular stabilization of the spine [32–37] provide a theoretical framework to investigate the mechanisms of chiropractic adjustments or spinal manipulation. The mechanical and physiologic influences of spinal manipulation on the targeted spinal tissues that have recently begun to be quantified experimentally represent an important first step in validating chiropractic theories. However, this work has been limited to animal models, noninvasive procedures, or minimally invasive procedures. For example, Pickar and McLain  measured afferent unit discharge to facet manipulation, and Pickar and Wheeler  measured muscle spindle and golgi-tendon organ responses to spinal manipulative-like loads in the feline. Basic animal research has now demonstrated the existence of neural discharge during spinal manipulative-like loads, but the results are not easily extrapolated in humans.
Intraoperative monitoring techniques have proven beneficial for monitoring neurophysiological events during spinal surgery, and such techniques have been used to study responses of spinal manipulation. Colloca et al  recently completed a pilot study investigating spinal nerve root action potential responses during intraoperative lumbosacral spinal manipulation. Spinal nerve root responses were found to be related to segmental contact point, and applied force vector and similarities were observed between internal and external thrusts. This study was limited to a single patient; nerve root measurements were unilateral; and the temporal relationships of the SMTs and nerve root responses could not be studied.
The purpose of the current study was to perform a comprehensive biomechanical and neurophysiological analysis of SMT in a series of 9 symptomatic patients. We hypothesized that neurophysiological and biomechanical responses would be related to the magnitude and location of the SMT, with differential responses dependent on patient symptomatology.
This clinical biomechanical study confirmed that spinal manipulation induces spinal motion and concomitant spinal nerve root responses. This line of investigation is the first to simultaneously measure vertebral movements and nerve root responses during SMT in human subjects. Such neuromechanical responses may be related to the mechanisms of spinal manipulation as administered in routine clinical practice.
Due to the invasiveness necessary to quantify spinal motions during spinal manipulation, previous research has typically been limited to cadaver studies. [1, 17, 18, 46] Gál et al  measured relative movements between vertebral bodies during PA thoracic SM. In this study, steel bone pins were embedded in the vertebral bodies of 2 unembalmed postrigor cadavers (aged 77 years each) at the levels of T10, T11, and T12. High-speed cinematography measured spinal motions during SM delivered at the level of T11. Preload and peak forces were approximately 80 N and 525 N, respectfully, in their study. These authors reported statistically significant mean relative translations and rotations ranged from 0.3 mm ± 0.2 mm to 0.6 ± 0.4 mm and 0.0 ± 0.3° to 1.9 ± 0.2°, respectively, between the 2 subjects. Similarly, Maigne and Guillon  measured relative lumbar spinal motions during lumbar spinal manipulation in 2 unembalmed cadavers (aged 49 and 71 years) by implanting accelerometers in the vertebral bodies. Using side-posture manipulation, the authors reported a maximum approximation between the L4–5 functional spinal unit of 1.1 mm, which is consistent with the magnitudes of relative vertebral movements observed in the current study. The ML, PA, and AX peak-to-peak vertebral displacements in this study are also of the same magnitude as previously reported in situ and in vivo relative or intervertebral motion studies.  Differences in the vertebral displacement response for the current study reflect subject differences, recording and sampling methodologies utilized, SMT force magnitude and duration, and segmental versus intersegmental nature of measurements.
Differences in vertebral motion responses associated with thrusts applied on various anatomical landmarks are important to clinicians who apply forces to the spine. In the current study, SMTs delivered to the FJs resulted in significantly (approximately 3–fold) greater ML motions as compared with SMTs delivered to the SPs. Because the SMT force vector was similar for thrusts on SPs and FJs, it is apparent that the segmental contact point has a direct influence on the vertebral motion response that is elicited. For clinicians, ML motion during spinal manipulation is accomplished by applying the SMT to the FJ as opposed to the SP. Moreover, in the case of the impulsive-type forces (force-time period « natural frequency) produced during MFMA SMT, the vertebral displacement response increased in a relatively linear manner with increasing force amplitude (constant preload).
A limitation of the current study was the fact that we did not quantify the precise thrust angle and FJ segmental contact points during the SMTs. Both of these factors may influence the motion response, but the surgical setting and the complexity of the motion and neurophysiological measurements performed precluded such measurements. Care was taken to perform the SMTs in a consistent and routine clinical manner, namely PA anterior-inferior or anterior-superior angulations of 20° ± 5° and offset of 10 mm to 15 mm from the midline (thrusts over FJs). Our aim was to quantify the lumbar vertebral motion response associated with spinal manipulation as it is performed in routine clinical chiropractic practice. According to computer simulations performed by Keller et al,  a 5° angulation difference (–15° versus –20°) and 5–mm contact point offset are predicted to result in less than a 0.1–mm difference in the peak-to-peak PA and axial motion responses to impulsive forces. Thus, lumbar spine PA and AX motion responses to impulsive forces are thought to be relatively insensitive to thrust angle/contact point variations of 20°/5 mm or less. While imaging technology is currently available to identify the underlying segmental contact points during biomechanical assessments, [10, 48] we do not believe that this specificity would have assisted our aim of quantifying vertebral motions during clinically applied SMT. Nevertheless, the influence of variations in precisely controlled force vector and contact point on the in vivo motion response deserves further consideration.
The MFMA instrument used for the SMTs produced a very short time duration (impulsive) force that induced a transient dynamic oscillatory motion response. For a given force amplitude, impulsive forces are associated with smaller displacements in comparison with longer duration, nonperiodic forces, such as those commonly applied during manual manipulation.  Consequently, high-precision, low-noise, dynamic accelerometers were used in this study to quantify the dynamic motion response of individual segments. The posterior-anterior, medial-lateral, and axial acceleration responses and displacements derived from the acceleration responses indicate that the method yields results comparable with other kinematic measurement methods, including spatial linkage sensors.  Additional work is needed to determine the reproducibility of the acceleration-based vertebral motion analysis method.
In the current study, we did not transform the Cartesian components of acceleration (x, y, z) to account for rotations of the vertebral segments or to estimate the flexion-extension rotation and medial-lateral rotation of the segments. Such transformations require knowledge of the location of the rotation axes relative to the accelerometer axes, and although we obtained fluoroscopic images of the pin-accelerometer sites, the image quality and image coverage was insufficient to perform these measurements in a manner precise enough to warrant transformation. Given the small absolute x, y, and z vertebral displacements measured (< 1 mm), vertebral rotations would be predicted to be extremely small, and therefore the transformed vertebral motions would not be expected to vary appreciably from that reported in this study. The absolute intervertebral flexion-extension rotations (< 1°) reported by Nathan and Keller  and vertebral and intervertebral flexion-extension rotations reported by Keller et al47 support this assumption. A 6–degree-of-freedom motion measurement system (3 translations and 3 rotations) would provide a more precise description of vertebral displacements and could be used to obtain vertebral rotations.
Our results are presented for patients undergoing surgery for significant spinal disorders and therefore should not be considered “normal lumbar segment motion responses.” As previously noted, investigations into spinal motions during spinal manipulation are in their infancy, so readily available data regarding spinal motions in normal subjects as opposed to subjects with spinal disorders are sparse.  A number of studies indicate that it is likely that spinal motions are highly dependent on the force-time input of the directed thrust, [14, 49, 50] as well as a variety of clinical factors, such as pain, [7, 13, 51] spinal morphology,  the presence of degeneration,16, 53, 54 and muscular stiffness. [55, 56] Therefore, vertebral motions observed in the spinal surgery patients are not expected to be representative of normal or asymptomatic subjects. Recent work by Kaigle et al  examined in vivo spinal motions and muscular responses in patients and asymptomatic subjects performing unresisted flexion-extension tasks. They found that intervertebral motions and trunk mobility were significantly lower in the patients than controls both in terms of range and pattern of motion. Still other factors such as intra-abdominal pressure,  cycle of breathing,  spinal level being tested, [22, 60] vector of applied force, [61–63], and spinal positioning during testing  have all been found to be important variables of spinal motion. In the current study, we accounted for many of these variables by placing patients in the same position on the same frame, standardizing the segmental level, vector, and cycle of breathing during performance of the SMTs. Further work in this regard with respect to understanding spinal motion differences among patients and asymptomatic subjects is warranted.
The results obtained from this study provide basic biomechanical information that is useful to both clinicians and researchers. The dynamic motion response data, force dependence, and coupling characteristics of the spinal segments to PA thrusts reported in this study will also assist researchers in the development and validation of computer models that aim to simulate the static and dynamic motion response of the spine. [47, 65–67]
Based on the results of this study, a recent model developed by Keller et al  is currently being refined to include motion coupling in each of the orthogonal axes of the spine.
Based on the knowledge of the presence of mechanosensitive afferents in the discoligamentous and muscular spinal tissues, we assumed that mechanical stimulation of viscoelastic structures during SMT would result in physiologic responses in human subjects. [25, 26, 29]
Prior research has demonstrated that mechanical and electrical stimulation of spinal articulations results in neurophysiological and neuromuscular responses, but such research has mostly been limited to the laboratory utilizing animal models. [36–39, 68] Intraoperative monitoring techniques are currently used in spinal surgery [69–73] and offer promise for evaluating neurophysiological responses during SMT,  albeit limited to the research setting. Thus, the objective of the current study was to measure intraoperative neuromechancial responses with a commonly used conservative therapeutic approach—spinal manipulation.
Because our measurements were taken just adjacent to the dorsal root ganglion, it is likely that the SMT-induced CAPs observed in the S1 spinal nerve roots were afferent traffic resulting from the stimulation of mechanosensitive afferent fibers in the viscoelastic spinal tissues. Sensory receptors within a tissue such as spinal ligaments, facets, disks, and muscles can initiate neural outflow to the spinal cord during application of various mechanical stimuli (eg, pressure, elongation, vibration, friction, tissue crushing) and application of chemical stimulants.  However, we were not able to directly ascertain the exact source of the neurophysiological responses, as is routinely performed in animal studies. [74, 75] Rather, intraoperative monitoring of compound action potentials was performed, which represents the algebraic sum of action potentials arising from respective mechanosensitive axons passing through the epineuria of the dorsal spinal nerve roots. Because the CAP represents many axons with differing thresholds of excitation, the CAP response is graded with a magnitude that is proportional to the intensity of stimulation.
We originally hypothesized that neurophysiological and biomechanical responses would be related to the magnitude and location of the SMT, with differential responses dependent on patient symptoms. Indeed, we found that variable intensity SMTs produced CAP responses of different amplitudes. Moreover, the magnitude of the CAP responses was significantly greater for SMTs compared with sham thrusts, indicating that the CAP response was not a product of preload. However, because we observed no difference in CAP response for MFMA SMTs delivered to the SPs or FJs, our findings indicate that spinal nerve root responses may not be sensitive to segmental contact point. Larger force magnitudes as delivered in other forms of manual SMT may cause more frequent and larger amplitude biomechanical and neurophysiological responses.  Further investigation into the effects of force-time profiles and segmental contact points on neuromechanical responses is warranted.
The mean reflexogenic time duration (SMT-to-peak positive CAP response) obtained in this study is similar to the work of others who have stimulated spinal structures and recorded physiological responses. [32, 33, 36, 69] Some researchers have used electrical stimulation to measure reflexogenic activity in the adjacent spinal musculature. Indahl et al [36, 68] reported time durations of 4 ms to 8 ms in a porcine model on stimulating the intervertebral disk and sacroiliac joint. Kang et al  also reported similar stimulus-to-response times of about 10 ms in feline preparations. Solomonow et al  measured stimulus-to-response time durations of 5 ms to 10 ms in human subjects on electrical stimulation of the supraspinous ligament. Stimulus-to-response times in the current study corroborate these time durations in our human subjects. It is likely that the CAP response represents afferent traffic from multiple mechanosensitive units in the muscular and discoligamentous soft tissues. The average 12–ms delay between the SMT and positive CAP response in the current study are expected due to the time it takes for the stimulus to travel along the Ia fibers, through the dorsal root ganglion, to the spinal cord. Neurologic deficits inherent in the patient population of the current study may have resulted in stimulus-to-response delays or the absence of positive CAP responses altogether. Indeed, a significant percentage of SMTs did not elicit positive neurophysiological responses in the patients. However, with the current methodology, it was not possible to ascertain whether the presence (or absence) and amplitude of CAP responses were specifically related to the neurologic status of the patient.
Nevertheless, it would not be unreasonable to expect neurologic deficits from damaged tissues. Three fourths of patients in this study had radiculopathy in the left lower extremity. Such clinical presentation might help to explain the greater number of right-sided (asymptomatic side) positive S1 CAP responses, as opposed to those measured from the left S1 spinal nerve root. This is consistent with the findings of Solomonow et al  who reported an absence of electromyography (EMG) responses during intraoperative stimulation of the supraspinous ligament. Hence, neurological deficit among patients may explain the decreased number of positive neurophysiological responses to SMT. In assessing the CAP response, positive responses were based on a threshold level of 2.5 × baseline. Responses at lower levels were not counted as “positive.” In a previous study,  peak-peak EMG reflex responses to PA thrusts were categorized according to 8 different baseline thresholds: >1.5×, >2.0×, >2.5×, >3.0×, >3.5×, >4.0×, >4.5×, and >5.0× the baseline p-p surface electromyography (sEMG) values. Here baseline refers to the resting or reference noise level of the biopotential (CAP in this study). A 1.5–fold increase (1.5×) represented a very weak reflex response, whereas a 5–fold increase (5.0×) represented a very strong reflex response. A 2.5× response was chosen for this study to ensure that the CAP responses were substantially greater than the background noise level. The clinical relevance of CAP threshold needs to be clarified further. A larger patient population will assist in clarifying the neuromechanical effects of SMT, including the effects of force vectoring, force-time profiles, and segmental contact points on neuromechanical responses. In particular, investigation of traditional manual SMT procedures  is necessary to better describe the neuromechanical responses of SMT.
Controversy may arise over our terminology reporting the use of “sham” SMT, since the so-called sham setting produces a 30 N peak impulse force. This setting has been referred to as a sham SMT by us and other investigators. [77, 78] Subsequently, both biomechanical and clinical studies have been performed using the zero (sham) and max settings of the device. Noteworthy, Keller and Colloca  found that the trunk muscle function assessed using erector spinae muscle electromyography was significantly improved in patients who received a max setting AAI SMT intervention. These authors found that there was no functional improvement in trunk muscle function for patients who received sham (0 setting) AAI SMTs or control (no intervention) treatment.
In the current study, the CAP response was temporally related to the onset of the MFMA SMTs and not to the initiation of the preload force. Although we did not include a control protocol that applied a preload force without engaging the AAI, our previous research showed that CAP responses were not elicited during the application of a preload force alone.  In this work, other control experiments, wherein the CAP electrode was intentionally moved on the spinal nerve root, were not found to produce a CAP response. Thus, we feel confident that the CAP responses observed in the current study are not experimental artifacts. From a data analysis point of view, engaging the AAI also helped to facilitate the neuromechanical temporal and amplitude measurements performed in this study.
Neurophysiologic models theorize that SMT may stimulate or modulate the somatosensory system and subsequently may evoke neuromuscular reflexes. [38, 79–81] Such reflexes are thought to inhibit hyperactive musculature, inhibit nociceptive traffic, and improve spinal function. The current line of investigation assists in understanding the relationships between the mechanical stimulation as delivered in SMT and the concomitant biomechanical and neurophysiological (neuromechanical) responses. In attempting to understand such neuromechanical relationships, often overlooked is the clinical status of the patient. The highly individualized neuromechanical response characteristics among patients in this study serves to highlight the need to clinically correlate the neuromechanical response characteristics with patient clinical status. Identifying such clinical relevance and understanding just how SMT may be related to inhibition or stimulation of the central nervous system in modulating nociception in humans awaits clarification. Our current work and the work of others aim to investigate such issues. [82–84]
In vivo PA impulsive force SMTs in human subjects were found to produce spinal nerve root responses that were temporally related to the onset of vertebral motion. These findings suggest that vertebral motions produced by spinal manipulation may play a prominent role in eliciting physiologic responses. Patient clinical status also appears to have a prominent role in the presence of neurophysiological responses. Further work, particularly examination of the force magnitude and frequency dependency of SMT, is necessary to elucidate the clinical relevance of enhanced or absent CAP responses in patients. Knowledge of biomechanical and neurophysiological events that occur during spinal manipulation assists in formulating a theoretical framework to understand the mechanisms of spinal manipulation.